Google launches advanced text-to-speech models for custom voice creation

Google has introduced two new text-to-speech models, Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS, designed to offer creators and developers highly expressive and customizable voice generation capabilities. These models enable the creation of realistic voices with precise control over accents, emotional tone, and dialogue delivery, supporting over 100 languages. The models are part of Google’s expanding Gemini Audio family, following previous releases like 3.5 Live Translate and 3.5 Transcribe. Gemini 3.8 Flash TTS has achieved the #1 spot on Hume AI’s Voice Design Benchmark and leads in accent modeling, while Flash-Lite TTS ranks #2 on the Overall Quality Index. Both models are rolling out today and are integrated into platforms like Google AI Studio, Gemini Notebook, and Google Vids. Google emphasizes safety measures, including consent verification for voice replication and SynthID watermarks to prevent misinformation. The new tools are being adopted by partners such as Figma, HeyGen, and Linguana to enhance global dubbing and localized media production. Developers can now build and deploy high-performance speech experiences using the Gemini API, with support for multilingual content and interactive voice agents.